Database Migrations - OpenClaw Mission Control

OpenClaw Mission Control uses Alembic for database schema migrations. This guide covers migration workflows, commands, and best practices.

Overview

Migrations are stored in backend/migrations/versions/ and manage schema evolution over time. The migration system:

Tracks applied migrations in the alembic_version table
Supports forward migrations (upgrade) and rollbacks (downgrade)
Auto-generates migration scripts from SQLModel schema changes
Validates migration graph consistency in CI

Migration Files

Directory Structure

backend/
├── alembic.ini              # Alembic configuration
├── migrations/
│   ├── env.py              # Alembic environment setup
│   ├── versions/           # Migration scripts
│   │   ├── 658dca8f4a11_init.py
│   │   ├── 99cd6df95f85_add_indexes_for_board_memory_task_.py
│   │   ├── a1e6b0d62f0c_drop_org_name_unique.py
│   │   ├── b4338be78eec_add_composite_indexes_for_task_listing.py
│   │   └── ...

`alembic.ini`

Configuration file at backend/alembic.ini:

[alembic]
script_location = migrations
prepend_sys_path = .
sqlalchemy.url = driver://user:pass@localhost/dbname  # Overridden by env.py

The sqlalchemy.url is dynamically set by migrations/env.py using the DATABASE_URL environment variable.

`migrations/env.py`

Runtime configuration that:

Loads app.models to discover SQLModel schemas
Reads DATABASE_URL from app.core.config.settings
Normalizes database URL (postgresql:// → postgresql+psycopg://)
Configures online/offline migration modes

Key function:

def get_url() -> str:
    """Return the normalized SQLAlchemy database URL for Alembic."""
    return _normalize_database_url(settings.database_url)

Running Migrations

Apply All Pending Migrations

Upgrade to the latest schema:

cd backend
.venv/bin/alembic upgrade head

Docker Compose:

docker compose exec backend alembic upgrade head

Makefile shortcut:

make backend-migrate

Apply Specific Migration

Upgrade to a specific revision:

cd backend
.venv/bin/alembic upgrade b4338be78eec  # Revision ID

Rollback Migrations

Downgrade to a previous revision:

cd backend
.venv/bin/alembic downgrade -1  # Go back one revision

Downgrade to a specific revision:

cd backend
.venv/bin/alembic downgrade a1e6b0d62f0c

Rollback all migrations (dangerous!):

cd backend
.venv/bin/alembic downgrade base

Rolling back to base will drop all tables. Only use this in development or when explicitly recreating the database.

Check Current Schema Version

cd backend
.venv/bin/alembic current

Output:

b4338be78eec (head)

View Migration History

cd backend
.venv/bin/alembic history --verbose

Automatic Migrations on Startup

The DB_AUTO_MIGRATE environment variable controls whether migrations run automatically when the backend starts:

# Enable auto-migration (useful in dev)
DB_AUTO_MIGRATE=true

# Disable auto-migration (recommended in production)
DB_AUTO_MIGRATE=false

Default behavior:

ENVIRONMENT=dev: DB_AUTO_MIGRATE defaults to true
All other environments: DB_AUTO_MIGRATE defaults to false

Production recommendation: Set DB_AUTO_MIGRATE=false and run migrations manually before deploying new code.

Creating New Migrations

Auto-Generate from Schema Changes

When you modify SQLModel models in backend/app/models/, generate a migration:

cd backend
.venv/bin/alembic revision --autogenerate -m "add user email field"

Alembic will:

Compare current database schema with SQLModel definitions
Generate a migration script in migrations/versions/
Detect added/removed tables, columns, indexes, and constraints

Example output:

INFO  [alembic.autogenerate.compare] Detected added column 'users.email'
Generating /path/to/backend/migrations/versions/abc123def456_add_user_email_field.py ... done

Manual Migration

Create an empty migration template:

cd backend
.venv/bin/alembic revision -m "custom data migration"

Edit the generated file in migrations/versions/ to add custom logic:

def upgrade() -> None:
    # Custom upgrade logic
    op.execute("""
        UPDATE tasks SET status = 'pending' WHERE status IS NULL
    """)

def downgrade() -> None:
    # Custom downgrade logic
    pass

Review Generated Migrations

Always review auto-generated migrations before applying:

def upgrade() -> None:
    # ### commands auto generated by Alembic ###
    op.add_column('users', sa.Column('email', sa.String(), nullable=True))
    op.create_index(op.f('ix_users_email'), 'users', ['email'], unique=True)
    # ### end Alembic commands ###

def downgrade() -> None:
    # ### commands auto generated by Alembic ###
    op.drop_index(op.f('ix_users_email'), table_name='users')
    op.drop_column('users', 'email')
    # ### end Alembic commands ###

Check for:

Unintended schema changes
Missing NOT NULL constraints
Reversibility of downgrade() function
Data migrations that need custom logic

Migration Validation (CI)

The Makefile includes a comprehensive migration validation target:

make backend-migration-check

This target:

Validates migration graph: Checks for cycles and orphaned revisions using scripts/check_migration_graph.py
Tests upgrade path: Spins up a temporary PostgreSQL container and applies all migrations
Tests downgrade path: Rolls back all migrations to base
Tests re-upgrade: Applies all migrations again to verify idempotency

Example CI workflow (from .github/workflows/ci.yml):

- name: Validate database migrations
  run: make backend-migration-check

Local Migration Testing

Test migrations against a clean database:

# Start a temporary Postgres container
docker run -d --rm --name mc-test-db \
  -e POSTGRES_PASSWORD=postgres \
  -e POSTGRES_DB=migration_test \
  -p 55432:5432 \
  postgres:16

# Wait for database to be ready
sleep 5

# Run migrations
cd backend
DATABASE_URL=postgresql+psycopg://postgres:postgres@localhost:55432/migration_test \
  .venv/bin/alembic upgrade head

# Test downgrade
DATABASE_URL=postgresql+psycopg://postgres:postgres@localhost:55432/migration_test \
  .venv/bin/alembic downgrade base

# Clean up
docker rm -f mc-test-db

Common Migration Operations

Add a Column

def upgrade() -> None:
    op.add_column('tasks', sa.Column('priority', sa.Integer(), nullable=True))

def downgrade() -> None:
    op.drop_column('tasks', 'priority')

Add an Index

def upgrade() -> None:
    op.create_index('ix_tasks_status', 'tasks', ['status'])

def downgrade() -> None:
    op.drop_index('ix_tasks_status', table_name='tasks')

Add a Composite Index

def upgrade() -> None:
    op.create_index(
        'ix_tasks_board_status',
        'tasks',
        ['board_id', 'status'],
        unique=False
    )

def downgrade() -> None:
    op.drop_index('ix_tasks_board_status', table_name='tasks')

Rename a Column

def upgrade() -> None:
    op.alter_column('tasks', 'old_name', new_column_name='new_name')

def downgrade() -> None:
    op.alter_column('tasks', 'new_name', new_column_name='old_name')

Data Migration

from alembic import op
import sqlalchemy as sa

def upgrade() -> None:
    # Add new column
    op.add_column('tasks', sa.Column('assignee_id', sa.String(), nullable=True))
    
    # Migrate data
    connection = op.get_bind()
    connection.execute(
        sa.text("""
            UPDATE tasks
            SET assignee_id = (SELECT user_id FROM assignments WHERE assignments.task_id = tasks.id LIMIT 1)
        """)
    )

def downgrade() -> None:
    op.drop_column('tasks', 'assignee_id')

Drop a Table (with Safety Check)

def upgrade() -> None:
    # Check if table is empty before dropping
    connection = op.get_bind()
    result = connection.execute(sa.text("SELECT COUNT(*) FROM deprecated_table"))
    if result.scalar() > 0:
        raise Exception("Cannot drop deprecated_table: contains data")
    
    op.drop_table('deprecated_table')

def downgrade() -> None:
    # Recreate table structure
    op.create_table(
        'deprecated_table',
        sa.Column('id', sa.Integer(), primary_key=True),
        sa.Column('name', sa.String(), nullable=False)
    )

Production Deployment Workflow

Pre-Deployment Migration

Review pending migrations:
```
cd backend
.venv/bin/alembic history
```

Backup database:

pg_dump -U postgres mission_control > backup_$(date +%Y%m%d_%H%M%S).sql

Apply migrations:
```
.venv/bin/alembic upgrade head
```

Verify migration:

.venv/bin/alembic current
psql -U postgres mission_control -c "\dt"  # List tables

Deploy application code:

systemctl --user restart mission-control-backend

Zero-Downtime Migrations

For large tables or production systems:

Make schema changes backward-compatible:
- Add nullable columns first, populate them, then add NOT NULL constraint
- Create new tables/indexes before dropping old ones

Use separate migration steps:

# Step 1: Add new column (nullable)
alembic revision --autogenerate -m "add email column nullable"
alembic upgrade head

# Step 2: Backfill data
alembic revision -m "backfill email data"
# Edit migration to populate email field
alembic upgrade head

# Step 3: Add NOT NULL constraint
alembic revision --autogenerate -m "make email not null"
alembic upgrade head

Test in staging before applying to production

Troubleshooting

Migration Failed Mid-Execution

Alembic may leave the database in an inconsistent state. Check the alembic_version table:

SELECT * FROM alembic_version;

If the version is incorrect:

# Manually set version (use with caution)
cd backend
.venv/bin/alembic stamp <correct_revision_id>

Migration Conflicts After Merge

If two branches created migrations in parallel, you may have multiple heads:

cd backend
.venv/bin/alembic heads

Merge the heads:

.venv/bin/alembic merge <revision1> <revision2> -m "merge migration branches"

“Can’t locate revision identified by ‘xxxxx’”

The migration file is missing from migrations/versions/. Either:

The migration was deleted (restore from git)
You’re on the wrong branch (switch to correct branch)
Your working directory is wrong (use cd backend)

Database URL Not Found

Ensure DATABASE_URL is set in backend/.env:

grep DATABASE_URL backend/.env

Or set it temporarily:

DATABASE_URL=postgresql+psycopg://postgres:postgres@localhost:5432/mission_control \
  .venv/bin/alembic upgrade head

Rollback Data Loss

Downgrade operations may lose data. Always:

Backup before downgrading
Review downgrade() logic in migration files
Test rollback in staging before production

Best Practices

Always review auto-generated migrations before committing
Test migrations locally before pushing to production
Backup production database before applying migrations
Keep migrations reversible when possible
Use descriptive migration messages: "add user email field" not "update schema"
Avoid editing applied migrations: Create a new migration instead
Run make backend-migration-check in CI to catch migration issues early
Coordinate with team when creating migrations to avoid conflicts

Reference

Alembic documentation: https://alembic.sqlalchemy.org/
SQLModel documentation: https://sqlmodel.tiangolo.com/
Migration history: backend/migrations/versions/
Environment configuration: backend/migrations/env.py
Settings class: backend/app/core/config.py

Get Started

Core Concepts

Guides

Deployment

Operations

​Overview

​Migration Files

​Directory Structure

​alembic.ini

​migrations/env.py

​Running Migrations

​Apply All Pending Migrations

​Apply Specific Migration

​Rollback Migrations

​Check Current Schema Version

​View Migration History

​Automatic Migrations on Startup

​Creating New Migrations

​Auto-Generate from Schema Changes

​Manual Migration

​Review Generated Migrations

​Migration Validation (CI)

​Local Migration Testing

​Common Migration Operations

​Add a Column

​Add an Index

​Add a Composite Index

​Rename a Column

​Data Migration

​Drop a Table (with Safety Check)

​Production Deployment Workflow

​Pre-Deployment Migration

​Zero-Downtime Migrations

​Troubleshooting

​Migration Failed Mid-Execution

​Migration Conflicts After Merge

​“Can’t locate revision identified by ‘xxxxx’”

​Database URL Not Found

​Rollback Data Loss

​Best Practices

​Reference

​Next Steps

Build docs developers (and LLMs) love

Overview

Migration Files

Directory Structure

`alembic.ini`

`migrations/env.py`

Running Migrations

Apply All Pending Migrations

Apply Specific Migration

Rollback Migrations

Check Current Schema Version

View Migration History

Automatic Migrations on Startup

Creating New Migrations

Auto-Generate from Schema Changes

Manual Migration

Review Generated Migrations

Migration Validation (CI)

Local Migration Testing

Common Migration Operations

Add a Column

Add an Index

Add a Composite Index

Rename a Column

Data Migration

Drop a Table (with Safety Check)

Production Deployment Workflow

Pre-Deployment Migration

Zero-Downtime Migrations

Troubleshooting

Migration Failed Mid-Execution

Migration Conflicts After Merge

“Can’t locate revision identified by ‘xxxxx’”

Database URL Not Found

Rollback Data Loss

Best Practices

Reference

Next Steps